Leaves optimization project is an open source optimization-based software platform. The project is further developed and maintained by scientists from Cardinal Operations (shanshu.ai) and the Research Institute of Interdisciplinary Sciences (RIIS) in Shanghai University of Finance and Economics (SHUFE) in China. The platform currently has three main components:

1. Mathematical Programming Solvers: Solvers for linear optimization, geometric programming, linearly constrained convex programming based on Python. Optimization solvers for various convex optimization (SOCP, SDP, etc) and non-convex optimization based on MATLAB/C are coming soon.

2. Machine Learning Toolkits: Efficient first-order algorithms are developed for various statistical and machine learning models, such as classification, regression, dimension reduction, etc. More sophisticated models in supervised and unsupervised learning, and deep learning, will be added soon.

3. Applied OR Software: Solution package for common OR problems including vehicle routing problem (VRP), inventory management, water network scheduling, automated guided vehicle (AGV) routing, etc.

The Leaves platform encourages researchers/enthusiasts in operations research, to contribute algorithms and implementations for solving optimization problems. We also encourage participants to submit your OR solutions to real word problems to be a part of the Leaves family. If you want to be part of the project, please contact us at leaves@shanshu.ai or rcmsia@mail.shufe.edu.cn. We thank you for your attention and passion.

  • Mathematical Programming

  • Machine Learning

    1. Machine Learning [ code ] [ manual ] [ dataset ]
  • Applied OR Software

    1. AGV Routing
    2. Efficient scheduling and routing algorithm is essential for advanced intelligent Automated Guided Vehicle System. Cardinal IP Algorithm which is developed all by Cardinal Scientist implements cutting-edge approximation algorithm to find the best scheduling plan in limited computational time. These techniques include the interior-point-method based LP solver, the problem structure dependent cutting-plan generation methods, the LP based global routing algorithm and the customized network flow algorithm. Integrating these methods, Cardinal IP algorithm has been successfully applied in many real AGV cases.
      [ manual ]
    1. StockGo
    2. StockGo is the first cloud intelligent software of inventory management, which based on SAAS. StockGo applys the state-of-arts mathematical models, combine real industry situation and evaluate the status of the inventory comprehensively. Without doubt, it is able to give suggestions of optimizations to users and help them to make decisions. As a result, StockGo can boost the response time and efficiency of supply chain management.
      [ link ]
    1. PonyPlus
    2. PonyPlus is a routing optimization system of logistics distribution owned by Cardinal Operations. Through our unique vehicle scheduling algorithm and techs of machine learning, it can analysis information of vehicles, orders and road conditions. Furthermore, following services: vehicle routing optimization, intelligent order matching and logistics distribution can provide a large variety of solutions to intelligent delivery routing planning. As result, PonyPlus can significantly increase efficiency of business operations
      [ link ][ demo ]